QumulusAI (QMLS) Investor presentation summary

Market opportunity and demand drivers

• AI infrastructure demand is rapidly increasing, with capacity as the main constraint.

• US datacenter power demand is projected to triple by 2030, reaching over 180 GW.

• Global datacenter build-out is expected to reach $7T by 2030.

• GenAI adoption has reached 53% of the global population in three years, outpacing previous tech waves.

• GPU rental prices are rising, with most new datacenter capacity pre-leased before completion.

Bottlenecks and solution approach

• Median grid interconnection wait times exceed five years, delaying new datacenter capacity.

• 30-50% of planned US datacenter capacity for 2026 is expected to slip or be canceled.

• QumulusAI deploys NVIDIA GPUs in colocation and sub-50MW data centers, bringing capacity online in months.

• The model emphasizes flexibility, access, cost efficiency, trust, and speed for clients.

Operations, network, and partnerships

• Operates six metro-adjacent colocation sites and two powered-land sites, with 39MW of grid power available.

• Pilot edge deployments at Internet Exchange Points (IXPs) aim to reduce latency and scale rapidly.

• Official NVIDIA Cloud Partner, validated for production-scale AI workloads.

• Partnerships include major tech and infrastructure firms such as NVIDIA, Dell, and Cisco.